Keywords

1 Introduction

Older adults rely more on walking for their daily journeys. However, as people age, their cognitive, perceptual and motor abilities decline. These abilities affect people navigating, as navigation requires people to perceive and integrate spatial information from the environment [4, 16]. Data showed that 53.58% of the lost population from 2016 to 2019 were elderly in China, and older adults had the higher risk of getting lost [3]. Therefore, it is essential to provide an adaptive navigation aid for older adults to avoid them getting lost.

With the development of mobile and location-based technology, mobile navigation aids have become one of the methods to help pedestrians navigate. In the fourth quarter of 2017, the users of using mobile navigation aid in China has reached 707 million and would continue to grow [7]. With the development of augmented reality technology, AR has been used in pedestrian navigation. Some studies have shown that AR could reduce the cognitive working load of the older adults in switching attention between the real world and information sources [9]. Mulloni et al. [13] believed that AR-based navigation systems were usually used at decision points, which could lead to its potential benefits by limiting the attention sharing of the elderly [17]. Considering to these advantages, AR-based pedestrian navigation aid may be beneficial to older adults. However, some study showed that older adults spent more time and had higher rate of navigation errors using AR-based navigation aid than using voice or digital map [12, 15]. In general, whether AR-based navigation aid is useful and usable for older adults still unknown. Therefore, the purpose of this study was aimed to investigate the usability and functionalities of AR-based pedestrian navigation aid.

2 Literature Review

Quantity of researches showed that navigation performance was influenced by individual characteristics, including age, gender, perceptual, and cognitive ability, which maybe benefit or hamper navigation [2, 10]. It has been shown in many studies that spatial ability was the major factor in determining performance. People with high spatial ability performed better than people with low spatial ability [2, 19]. However, as people aging, people’s spatial ability declined and needed more time to navigate [16]. Three types of spatial knowledge were required during navigating in a new environment: landmark knowledge, route knowledge and configuralknowledge [18]. Older adults needed more time to learn and elaborate spatial knowledge than young adults, and they had problems in use of configural knowledge [8, 16]. Therefore, using overview map could not improve their navigation performance. Even though, older adults were likely to use overview map, as people felt secure [16] and ensured that they are on the right direction or path by using map.

As a new navigation method, AR-based navigation aid superimposes position and orientation information in real world. In this way, user could not be distracted when they are always staring at the screen [1]. However, AR-based navigation had worse navigation performance, higher cognitive working load and lower system usability comparing with other type of navigation aids, such as voice and digital map [12, 15]. One reason possibly was the participants had no previous experience with AR technology [6]. In addition, the discontinuous and inaccurate tracking of AR-based navigation aid could make users make wrong decisions at intersections [13]. The design of virtual navigation objects, such as their color, also affected user performance and experience, as brightness and screen-glare affected the visibility of virtual navigation objects when they were superimposed on the real world [5, 12].

3 Methodology

3.1 Equipment and Material

Three criteria were used to choose the AR-based navigation aid application: the application should have overview map and text instruction, which are essential functions for users; the application should be stable to use; the application should be accessible. Finally, Baidu Map application with version 10.15.0.915 which has AR-based navigation function was used in this study. Baidu Map APP is a commercial application developed by Baidu to provide navigation aid for pedestrians and drivers and has the highest usage rate in Chinese navigation aid application market.

The interface of AR-based navigation aid consists of three main functions: virtual route, virtual navigator and auxiliary map. Virtual route is a blue line and the basic function of this tool. Virtual navigator is a cartoon character that users can choose to use it if they want. Auxiliary map is an overview map that shows user’s current location, destination and surrounding landmark. Figure 1 is the interfaces of the AR-based navigation aid we used in this study.

Fig. 1.
figure 1

Screenshots of AR-based navigation aid interface. (a) pedestrian was guided only by virtual route. (b) pedestrian was guided by virtual route and virtual navigator. (c) pedestrian was guided by virtual route, virtual navigator and auxiliary map. (d) pedestrian was guided by virtual route and auxiliary map.

Full size image

In this experiment, we used a smart phone, Huawei honor 8 with Android 6.0, to install the application. A camera, Sony FDR-AX40, was used to record the process.

3.2 Variables

The independent variables were virtual navigator presentation and auxiliary map presentation, both of which were within-subject variables. Virtual navigator presentation had two levels: with and without virtual navigator in AR-based navigation aid interface. The auxiliary map also had two levels: with and without auxiliary map in the navigation aid interface. Therefore, there were four types of navigation aid interfaces in this study, as shown in Fig. 1.

The dependent variables were user performance and satisfaction. There were three measures of user performance: success rate, navigation error and speed variation. Successful task was considered as participants competed tasks without any help of experimenter. A navigation error was participants deviated from the right route indicated by the navigation aid. Speed variation was defined as the difference between speed of participants using AR-based navigation aid minus the speed of without using navigation aid.

Covariates included demographic variables and road complexity. The demographic variables included: age, prefer walking speed (PWS). Road complexity was related to the length of road where participants passed, the number of intersections and the number of branch routes at each intersection. the formula of calculating road complexity was shown as (1).

$$ road \;complexity = \frac{{\mathop \sum \nolimits_{i = 1}^{n} log_{2} \left( {\text{Fi}} \right)}}{\text{L}} $$
(1)

Note: L is the length of whole road; n represents the number of intersections in the road; Fi represents the number of branch routes at each intersection.

3.3 Participants and Tasks

To examine age difference, a total of 28 participants, 14 young adults (M = 23.8, SD = 0.939, Rang = 22–25) and 14 older adults (M = 67.6, SD = 4.484, Rang = 61–76) were recruited in this study. Age groups were gender-balanced, with 7 females and 7 males in each group. All young adults were students from Chongqing University, and older adults were recruited from campus and communities near the college. Disabled subjects were excluded from this experiment. Only one young adult had the experience of using AR-based navigation aid before. Mostly participants (26 out of 28) had the experience of using technology products (computer and smart phone). Characteristics of participants were shown in Table 1.

Table 1. Demographic characteristics of participants
Full size table

Four different routes were predefined in this study, and each participant needed to navigate in the four routes by using different AR-based navigation aid interface, as shown in Table 2. As some of the participants were familiar with the campus, we did not tell them the destination of each routes and emphasized that they should strictly follow the instructions of the navigation aid.

Table 2. Details of four routes
Full size table

3.4 Procedure

The experiment required approximately 50 min. First, experimenter briefly introduced the experiment procedure and asked participants to complete a basic questionnaire about their background and the experience in using navigation aid. Second, we tested the PWS of each participant and showed participants a demonstration of think-aloud. In formal experiment, participants were asked to navigate in 4 routes by using AR-based navigation aid, while experimenter recorded participant’s thought and navigation process by using camera. Then, participants needed to complete a Post Study System Usability Questionnaire (PSSUQ) which used a five-point Likert scale and was carefully translated into Chinese [14]. Finally, a brief interview was conducted to collect participant’ s attitudes towards the AR-based navigation aid.

4 Results and Discussion

4.1 Performance

Success Rate.

Results showed that almost all participants could reach to the destinations by using AR-based navigation aid, with a success rate of 97.32%. Only two older adults failed to reach the destination in three navigation tasks. One of them navigated to an air-raid shelter where was no GPS signal, and said she did not know what to do next. Another participant made multiple navigation errors and finally abandoned the experiment.

Navigation Error.

The average number of navigation errors was 1.08 (SD = 1.23). Older adults made more errors (M = 1.42, SD = 1.564) than young adults (M = 0.79, SD = 0.801). However, no significant difference was observed between older adults and young adults.

The data was analyzed by using two-way repeated ANOVA model. Three covariates were included in the model. Prior to the analysis, we performed a regression line homogeneity test. Significant interaction effect (F = 68.9083, p < 0.01) between road complexity and virtual navigator presentation was observed. Therefore, we took road complexity as an independent variable and divided it into two levels: simple and complex.

The results showed that there was a main effect of virtual navigator presentation on navigation error (F = 4.538, p < 0.05), as shown in Fig. 2 (a). Participants had higher navigation error when virtual navigator was presented than when virtual navigator was not presented in navigation aid interface. Older adults had more navigation errors when presenting or not presenting virtual navigator than young adults, while the difference between them was not statistically significant. Possible reasons are that processing the navigation information provided by the virtual navigator required higher perceptual speed which is declined with age [11] and presenting the virtual navigator increased the cognitive working load of the participants.

Fig. 2.
figure 2

(a) Influence of virtual navigator on navigation error. (b) Influence of auxiliary map on navigation error. (c) Interactive effect between virtual navigator and road condition on navigation error.

Full size image

A main effect of auxiliary map presentation on navigation error was also observed (F = 28.128, p < 0.01). Presenting the auxiliary map made the number of errors increased by 61.53%. Older adults showed significant difference between navigating with auxiliary map and without auxiliary map (F = 20.54, p < 0.01), while no significant difference was fond on young adults. The possible reasons may be related to user’s cognitive working load and spatial ability. When the auxiliary map was presented, participants needed to match it with real route, virtual routes, which increased their cognitive working load. In addition, older adults had difficulty using overview map, for their declined spatial ability [16]. What is more, young adults were familiar with overview map, as they used it in their daily life.

Interactive effect between virtual navigator and road condition on navigation errors was observed (F = 53.112, p < 0.05), as shown in Fig. 2 (c). In simple road condition, presenting virtual navigator could reduce navigation errors (F = 30.61, p < 0.01). However, in complex road condition, presenting virtual navigator would significantly increase navigation error (F = 53.78, p < 0.01). In addition, we fond navigation errors often occurred in where road conditions were complex (e.g., an intersection with four branches or roundabout). One possible reason is that presenting virtual navigator in complex road required more attention from participants and increased their cognitive working load.

Speed Variation.

Speed variation was used to measure the degree of hesitation of participants during using AR-based navigation aid. There was no significant difference between virtual navigator and auxiliary map on speed variation. Participants seldom stopped during navigation tasks. When unsure which branch to choose at an intersection, participants would continue to walk and choose one of branch routes to try. Even if they made an error, they could quickly start navigating again. However, the covariate of PWS had a significant difference (F = 20.1992, p < 0.01) on speed variation. The bigger the PWS, the smaller the speed variation.

4.2 Satisfaction

The PSSUQ and interview were analyzed. The average PSSUQ score of all participants on AR-based navigation aid was 78.7 (SD = 19.81), and there was a significant difference between older and young adults (F = 16.48, p < 0.01), as shown in Fig. 3. Although older adults had more errors in navigation tasks, they had higher positive feedback (M = 90.81, SD = 10.46) than young adults (M = 66.58, SD = 19.74). During the interview, participants were asked their general feeling about the aid and problems they encountered. Most of older adults thought the aid was easy to learn and use than other types of navigation aids, such as paper map. However, young adults said they preferred to use 2D or 3D digital map rather than AR-based navigation aid, as they always got confused when the virtual route did not match the real route and they thought the auxiliary map was too small to see.

Fig. 3.
figure 3

The effect of age group on PSSUQ score

Full size image

What is more, we fond older and young adults had different opinions towards auxiliary map. Older adults said they seldom used auxiliary map during navigating because they were unfamiliar with overview map, and the map was too small and complex for them. However, young adults thought auxiliary map played a critical role during navigating, as map could help them know the general direction of destination.

4.3 List of Usability Problems

Through the analysis of the experimental video and the interviews, 11 usability problems were collected in this study (see Table 3). The number of participants who encountered the problems and total frequency of occurrence were countered. Usability problems were determined according to the following criteria: 1) participants expressed some negative effects, or treated something as problems; 2) participants felt confused; 3) participants made mistakes; 4) participants stopped the navigation tasks. Through statistical analysis, we found that the change delay of virtual route instruction is the most frequency problem participants encountered, which caused participants to deviate from the correct route or stop. In addition, most of participants were confused by the instructions of virtual navigator, virtual route and voice, especially at intersections.

Table 3. List of usability problems with AR-based navigation aid
Full size table

5 Conclusion

This study explored the usability problems of AR-based navigation aid in a field experiment. Participants’ performance and satisfaction were investigated. In general, participants could successfully reach to destinations by using AR-based navigation aid. Therefore, it is useful to use AR-based navigation aid. However, despite the high success rate, participants still made many navigation errors, especially older adults. In addition, presenting virtual navigator and auxiliary map would increase the navigation errors, so we suggested that it is better to avoid the use of virtual navigator and auxiliary map for older adults, as too many instruction information would increase their cognitive working load. For young adults, it is necessary to present auxiliary map for them to ensure their sense of control.

From this study, we collected some of usability problems. AR-based navigation aid instruction needs to be more responsive at intersections to avoid users deviating from the correct route or waiting for the change of instructions. In addition, participants could hardly match the instructions of virtual route and virtual navigator with real route, especially when there were multiple branches in the same direction at an intersection. Therefore, it is important to consider the design of AR-based navigation aid at complex intersections.